Comparative Genomics of Rumen Butyrivibrio spp. Uncovers aContinuum of Polysaccharide-Degrading Capabilities

Nikola Palevich,a William J. Kelly,b Sinead C. Leahy,a Stuart Denman,c Eric Altermann,a Jasna Rakonjac,d

Graeme T. Attwooda

aAgResearch Limited, Grasslands Research Centre, Palmerston North, New ZealandbDonvis Limited, Palmerston North, New ZealandcAgriculture and Food (CSIRO), St. Lucia, Queensland, AustraliadInstitute of Fundamental Sciences, Massey University, Palmerston North, New Zealand

ABSTRACT Plant polysaccharide breakdown by microbes in the rumen is funda-mental to digestion in ruminant livestock. Bacterial species belonging to the rumengenera Butyrivibrio and Pseudobutyrivibrio are important degraders and utilizers oflignocellulosic plant material. These bacteria degrade polysaccharides and fermentthe released monosaccharides to yield short-chain fatty acids that are used by theruminant for growth and the production of meat, milk, and fiber products. Althoughrumen Butyrivibrio and Pseudobutyrivibrio species are regarded as common rumen in-habitants, their polysaccharide-degrading and carbohydrate-utilizing enzymes arenot well understood. In this study, we analyzed the genomes of 40 Butyrivibrio and6 Pseudobutyrivibrio strains isolated from the plant-adherent fraction of New Zealanddairy cows to explore the polysaccharide-degrading potential of these important ru-men bacteria. Comparative genome analyses combined with phylogenetic analysis oftheir 16S rRNA genes and short-chain fatty acid production patterns provide insightinto the genomic diversity and physiology of these bacteria and divide Butyrivibriointo 3 species clusters. Rumen Butyrivibrio bacteria were found to encode a largeand diverse spectrum of degradative carbohydrate-active enzymes (CAZymes) andbinding proteins. In total, 4,421 glycoside hydrolases (GHs), 1,283 carbohydrate es-terases (CEs), 110 polysaccharide lyases (PLs), 3,605 glycosyltransferases (GTs), and1,706 carbohydrate-binding protein modules (CBM) with predicted activities involvedin the depolymerization and transport of the insoluble plant polysaccharides wereidentified. Butyrivibrio genomes had similar patterns of CAZyme families but variedgreatly in the number of genes within each category in the Carbohydrate-Active En-zymes database (CAZy), suggesting some level of functional redundancy. These re-sults suggest that rumen Butyrivibrio species occupy similar niches but apply differ-ent degradation strategies to be able to coexist in the rumen.

IMPORTANCE Feeding a global population of 8 billion people and climate changeare the primary challenges facing agriculture today. Ruminant livestock are impor-tant food-producing animals, and maximizing their productivity requires an under-standing of their digestive systems and the roles played by rumen microbes in plantpolysaccharide degradation. Members of the genera Butyrivibrio and Pseudobutyriv-ibrio are a phylogenetically diverse group of bacteria and are commonly found inthe rumen, where they are a substantial source of polysaccharide-degrading en-zymes for the depolymerization of lignocellulosic material. Our findings have high-lighted the immense enzymatic machinery of Butyrivibrio and Pseudobutyrivibrio spe-cies for the degradation of plant fiber, suggesting that these bacteria occupy similarniches but apply different degradation strategies in order to coexist in the competi-tive rumen environment.

Citation Palevich N, Kelly WJ, Leahy SC,Denman S, Altermann E, Rakonjac J, AttwoodGT. 2020. Comparative genomics of rumenButyrivibrio spp. uncovers a continuum ofpolysaccharide-degrading capabilities. ApplEnviron Microbiol 86:e01993-19. https://doi.org/10.1128/AEM.01993-19.

Editor Charles M. Dozois, INRS—InstitutArmand-Frappier

Copyright © 2019 Palevich et al. This is anopen-access article distributed under the termsof the Creative Commons Attribution 4.0International license.

Address correspondence to Graeme T.Attwood, graeme.attwood@agresearch.co.nz.

Received 3 September 2019Accepted 16 October 2019

Accepted manuscript posted online 25October 2019Published

EVOLUTIONARY AND GENOMIC MICROBIOLOGY

crossm

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 1Applied and Environmental Microbiology

13 December 2019

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from

KEYWORDS rumen, bacteria, polysaccharide, Butyrivibrio, Pseudobutyrivibrio, genome,CAZy, enolase

The need to feed a growing global population (1) is driving renewed interest inunderstanding the role of the rumen microbiota in the degradation and conversion

of plant polysaccharides into high-value animal products (2). The rumen is one of themost efficient plant polysaccharide depolymerization and utilization systems known,and its microbes are promising sources of fibrolytic enzymes for application in theproduction of biofuels from lignocellulosic material (3). Rumen bacteria are responsiblefor most of the breakdown of plant fiber via close interactions among phylogeneticallydifferent, but physiologically complementary, bacterial species (4, 5). Species belongingto the genera Butyrivibrio and Pseudobutyrivibrio form a significant group of rumenbacteria (6, 7) and are among a small number of rumen microbes capable of utilizingxylans and pectins (8–13). Butyrivibrio species contribute to fiber digestion in bothanimals (14–17) and humans (18) due to their ability to degrade hemicelluloses (19–22)and are also involved in protein breakdown (23) and the biohydrogenation of fattyacids (24, 25). At present, the genus Butyrivibrio includes the rumen species Butyrivibriofibrisolvens, B. hungatei, and B. proteoclasticus and the human species B. crossotus(26–30), while the genus Pseudobutyrivibrio has two species, Pseudobutyrivibrio xylaniv-orans and P. ruminis. Due to the substantial morphological (31), metabolic (32–34), andserological (35, 36) differences, it is likely that more distinct species groups of Butyriv-ibrio and Pseudobutyrivibrio exist in the rumen.

Butyrivibrio and Pseudobutyrivibrio strains encode a more impressive repertoire ofcarbohydrate-active enzymes (CAZymes) than most Firmicutes (7), including thoseinvolved in the degradation of pectin (glycoside hydrolase 28 [GH28], polysaccharidelyase 1 [PL1], PL9, PL10, PL11, carbohydrate esterase 8 [CE8], CE12) and xylan (GH8,GH10, GH11, GH43, GH51, GH67, GH115, GH120, GH127, CE1, CE2) (7, 37). Here, weprovide a multistrain systematic phenotypic and comparative genomic analysis ofrumen Butyrivibrio and Pseudobutyrivibrio species and show that they are capable ofgrowing on a range of carbohydrates, from simple mono- or oligosaccharides tocomplex plant polysaccharides, such as pectins, mannans, starch, and hemicelluloses.

(This research was conducted by N. Palevich in partial fulfillment of the requirementsfor a Ph.D. from Massey University, Manawatu, New Zealand, 2016 [37].)

RESULTSRumen Butyrivibrio strains are phylogenetically diverse. Phenotypic character-

izations, including the characterization of cell morphology, motility, carbon sourceutilization, and fermentation end products, and genotypic characterizations, includingcharacterization by 16S rRNA gene sequencing and pulsed-field gel electrophoresis(PFGE), were carried out on 30 Butyrivibrio strains from the rumen environment.Microscopic evaluation of cells from liquid cultures and from colonies on platesconfirmed that each of the 30 Butyrivibrio strains displayed morphologies consistentwith those of Butyrivibrio strains (see Data Set S1 in the supplemental material). Basedon analysis of full-length 16S rRNA gene sequences (Fig. S1), all Butyrivibrio strainsclustered separately from Pseudobutyrivibrio strains and grouped into three clusters.Cluster 1 contained the sequences of the type strains of B. proteoclasticus (B316T) andB. hungatei (JK615T) and 10 other Butyrivibrio strains. Cluster 2 contained the sequencesof 12 Butyrivibrio strains, none of which were type strains, and cluster 3 consisted of thesequences of 8 strains containing the B. fibrisolvens type strain (D1T) and closely relatedstrains (Fig. S1). Clusters 2 and 3 were well supported by bootstrap analyses, showing92% and 97% bootstrap support, respectively, while cluster 1 was more diverse,showing only 57% bootstrap support (Data Set S1). These results suggest that theButyrivibrio 16S rRNA gene sequences can be divided into two relatively cohesiveclusters, clusters 2 and 3, while the larger cluster, cluster 1, is a continuum of relatedsequences containing those of at least two species.

Palevich et al. Applied and Environmental Microbiology

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 2

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from

PFGE analyses of genomic DNAs digested with the restriction endonucleases (REs)ApaI and I-CeuI produced unique banding patterns for all of the Butyrivibrio strainsanalyzed, providing evidence for differences at the genomic DNA level between theseorganisms. The genome size estimates from the RE digests ranged from approximately3.5 Mb to 5.6 Mb (Data Set S1), with the average size being 4.14 Mb. PFGE analyses ofundigested genomic DNAs also identified large extrachromosomal elements, with anaverage size range of from 300 to 500 kb. The largest extrachromosomal DNA was an869.2-kb element from Butyrivibrio sp. strain XPD2002, and the smallest was a 99.1-kbelement from Butyrivibrio sp. strain AE3006 (Data Set S1). Comparisons of the draftgenomes in regard to the sizes of the extrachromosomal elements identified that twoof the six Butyrivibrio sp. strain AE3004 contigs matched the 433.1- and 350.1-kb bandsobserved. B. fibrisolvens FE2007 and WTE3004, as well as Butyrivibrio sp. strains AE3006,MB2005, AC2005, LC3010, VCD2006, WCD2001, VCB2006, XBB1001, FCS006, andNC2007, also possessed contigs similar to the PFGE bands observed. Overall, thepatterns from the PFGE analysis indicate that extrachromosomal elements are a com-mon genomic characteristic of rumen Butyrivibrio species.

Comparative genome analyses were carried out on the 30 Butyrivibrio strains, alongwith an additional 10 Butyrivibrio and 6 Pseudobutyrivibrio strains (Data Set S1), usingfunctional genome distribution (FGD), average nucleotide identity (ANI), and alignmentfraction (AF) analyses (38, 39). The findings of FGD analysis (Fig. 1) were consistent withthe phylogenetic inferences based on the full-length 16S rRNA gene sequence data(Fig. S1). The ANI and AF identities of whole-genome nucleotide sequences between

FIG 1 FGD of Butyrivibrio (B.) and Pseudobutyrivibrio (P.) genomes. The predicted ORFeomes of all 46 genomes were subjected to an FGD analysis, and theresulting distance matrix was imported into MEGA6 (82). The functional distribution was visualized using the UPGMA method (113, 114). The tree is drawn toscale, with the branch lengths being in the same units as those of the functional distances used to infer the distribution tree. The bar represents the numberof nucleotide substitutions per site.

Polysaccharide-Degrading Capabilities of Butyrivibrio Applied and Environmental Microbiology

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 3

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from

the proposed clusters of rumen Butyrivibrio species varied considerably between thegenome pairs (Data Set S1). The ANI and AF identities of the 6 Pseudobutyrivibrio strainscompared with those of the 40 Butyrivibrio strains were between 70% and 71% and�0.2, respectively. The clustering of the Butyrivibrio and Pseudobutyrivibrio genomesbased on Pfams, COGs, TIGRfams, and KO protein/functional family types (Data Set S1)were also generally consistent with the 16S rRNA gene-based species grouping andgenome similarity comparisons.

Although the demarcation between the Butyrivibrio and Pseudobutyrivibrio generawas distinct, the boundaries between species within each genus were less distinct. Ineach of the three Butyrivibrio clusters, some strains formed clearly separate groupings.For example, Butyrivibrio sp. strains MC2013 and NC3005 clustered away from the restof the cluster 3 B. fibrisolvens strains, and Butyrivibrio sp. strains AE3003, AE2005, andLB2008 and B. hungatei NK4A153 were distinct within Butyrivibrio cluster 1 (Fig. 1).

Butyrivibrio genomes include a large number of orthologous gene families. Thecore, variable, and unique gene families present in the Butyrivibrio (clusters 1 to 3) andPseudobutyrivibrio genomes were determined using BLAST analyses. A total of 29,105orthologous gene families were found (Fig. 2 and S2), of which 602 represented thegene families shared among all genomes, or the core genome set. The core genome setconsisted mainly of genes encoding housekeeping, carbohydrate metabolism, andtransport functions. The Pseudobutyrivibrio genomes had the highest number of uniquegenes (n � 471), with predicted functions including flagellum biosynthesis, signaltransduction, and the production of acetolactate synthase (ALS), accessory Sec systemproteins, diguanylate cyclase (GGDEF), phosphoesterase, cell division protein (FtsA),helicase, �-glucosidase, and GH3 proteins.

Genome alignments using the B. proteoclasticus B316T genome (chromosome,chromid, and plasmids) as the reference (Fig. S3) revealed that only Butyrivibrio sp.strain XBB1001, Butyrivibrio sp. strain VCB2006, and B. proteoclasticus FD2007 fromcluster 1 had significant synteny to B316T. FD2007 was the only genome which has achromid similar to that of the B316T genome. The high similarity of XBB1001, VCB2006,and FD2007 to B316T in the FGD, ANI, and protein family analyses supports theirclassification as strains of B. proteoclasticus.

The amino acid usgae and codon usage within the predicted proteomes of Butyr-ivibrio and Pseudobutyrivibrio were compared (Fig. S4). Isoleucine, leucine, serine,aspartate, glutamate, alanine, lysine, valine, and glycine were the most frequently usedamino acids. Codon usage was similar among the Pseudobutyrivibrio proteomes, whilecodon usage varied across the Butyrivibrio strains. These findings on codon usage areconsistent with those seen among the Firmicutes.

Butyrivibrio species are capable of using a wide range of polysaccharides.Substrate utilization tests and analyses of fermentation end products were carried outto determine the metabolic capacities of the Butyrivibrio strains evaluated in this study.There were clear patterns of substrate use by the species groups (Data Set S1). Moststrains were able to use glucose, galactose, melezitose, trehalose, glycerol, myoinositol,mannitol, and sorbitol. Cluster 2 and 3 strains were more versatile than cluster 1 strainsin terms of soluble carbohydrate utilization. Most Butyrivibrio strains were able to useall the substrates tested, apart from arabinose, mannose, rhamnose, maltose, melibiose,xylitol, amygdalin, esculin, rutin, and salicin. Many of the cluster 3 strains and some ofthe cluster 1 strains were also able to grow on the semisoluble or insoluble substratespectin and xylan, while none of the strains could use ball-milled cellulose. The B.fibrisolvens MD2001, AB2020, FE2007, ND3005, YRB2005, and WTE3004 strains displayedsimilar growth patterns, in particular, the ability to utilize pectin, xylan, and xylose,whereas Butyrivibrio sp. strains MC2013 and NC3005 expressed different growth pat-terns that coincided with their phylogenetic and genomic divergence within cluster 3(Fig. 1). All cluster 2 strains could utilize xylose, and strains AE3004, XPD2002, andVCD2006 displayed similar growth patterns on dextrin, inulin, starch, and xylan. Inter-estingly, cluster 2 strains were not able to utilize pectin for growth. The cluster 1 strainsB316T, VCB2006, XBB1001, and AE3009, which grouped closely at the genome and

Palevich et al. Applied and Environmental Microbiology

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 4

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from

phylogenetic levels, displayed similar growth patterns across all insoluble substratesanalyzed, and only AE2015 and B316T could utilize xylose (Data Set S1).

The polysaccharide-degrading capabilities encoded by the Butyrivibrio genomeswere further defined using CAZyme analysis. A total of 159 CAZyme families wereidentified (Fig. 3), consisting of 64 glycoside hydrolases (GHs), 14 carbohydrate es-terases (CEs), 7 polysaccharide lyases (PLs), 38 carbohydrate-binding protein modules(CBM), and 36 glycosyltransferases (GTs) (Data Set S2). Within the Butyrivibrio species,the strains generally had similar types of CAZymes, but the absolute number of geneswithin each of their categories in the Carbohydrate-Active Enzymes database (CAZy)varied considerably (Fig. 4; Data Set S2). In particular, B. fibrisolvens strains MD2001,AB2020, YRB2005, and WTE3004 from cluster 3 and cluster 1 strains MC2021, FD2007,VCB2006, XBB1001, FCS006, and AE3009 possessed the largest number of CAZymeswithin their respective groups (Fig. 4) and grouped together, based on the relativeabundance of CAZymes (Fig. 5). Interestingly, the CAZyme cluster analysis indicatedthat Butyrivibrio strain AE3003 and B. hungatei NK4A153, AE2005, and LB2008 clusteredmost closely with Pseudobutyrivibrio strains and well away from their nearest phyloge-

FIG 2 Flower plot diagram of unique, group-specific, and core gene families in the Butyrivibrio and Pseudobutyrivibrio genomes. The core genome is shown inthe center circle. Each colored segment represents the number of gene families shared among the four species groups, and the outer petals represent uniquegene families for individual genomes.

Polysaccharide-Degrading Capabilities of Butyrivibrio Applied and Environmental Microbiology

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 5

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from

FIG 3 Distribution of each CAZyme class and family in Butyrivibrio and Pseudobutyrivibrio genomes. Colored bars represent thetotal numbers of genomes that contain members of the specific CAZyme family present in their genomes.

Palevich et al. Applied and Environmental Microbiology

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 6

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from

netic relatives, B. hungatei MB2003 and B. proteoclasticus. Strains MC2013 and VCD2006and, to a lesser extent, strain NC2002, also had CAZyme profiles atypical of those oftheir closest relatives and were separated by CAZyme analysis.

Pfam domain analysis of the most abundant GH families (GH2, GH31, GH3, GH13,and GH43) showed that most did not contain signal sequences and, hence, werepredicted to be located intracellularly. Similarly, CAZymes with predicted roles in xylanand pectin degradation (the GH8, GH28, GH39, GH51, GH67, GH88, GH105, GH115, CE2,and CE10 families) were also predicted to be intracellular (Data Set S2), suggesting thata variety of complex oligosaccharides resulting from extracellular hydrolysis are trans-ported and metabolized within the cell.

Fermentation pathways and enolase gene loss. The fermentation pathways inButyrivibrio predicted from gene content and metabolic pathway reconstruction are shown

FIG 4 Comparative analysis of annotated Butyrivibrio and Pseudobutyrivibrio CAZymes. The numbers and types ofCAZyme modules or domains are represented as colored horizontal bars.

Polysaccharide-Degrading Capabilities of Butyrivibrio Applied and Environmental Microbiology

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 7

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from

in Fig. 6. Genome analysis and metabolic pathway reconstruction of Butyrivibrio strainsrevealed that 7 of the 8 cluster 3 strains (all except MC2013), cluster 2 strains WCD2001,VCD2006, AC2005, FC2001, and XPD2002, and cluster 1 strain AE2015 had all of the genesencoding the enzymes required for fermenting hexoses through to pyruvate via an intactEmbden-Meyerhof-Parnas (EMP) pathway. However, 18 Butyrivibrio genomes lack an iden-tifiable enolase gene (eno), which encodes the enzymatic conversion of 2-phospho-D-glycerate to phosphoenol pyruvate (EC 4.2.1.11) in the second-to-last step of the EMPpathway. Because these genomes are not fully closed, it is possible that the unsequencedregions contain the missing eno genes; therefore, genomic DNA from each strain wasscreened in PCRs using primers specific for eno genes. The PCR screens produced an enoamplicon in 8 Butyrivibrio strains but failed to produce an amplicon in 24 strains (19putatively eno-negative strains and 5 predicted eno-positive strains [strains NC3005,WCD2001, VCD2006, AC2005, and AE2015]) (Data Set S2). All eno-positive, PCR-positivestrains had strong alignments of both the forward and reverse primer sequences with thesequences of their encoded eno genes (Fig. S5); however, the 5 eno-positive, PCR-negativestrains did not. Analysis of the Eno_N (PF03952) and Eno_C (PF00113) Pfam domains of thepredicted enolase proteins showed that all cluster 3 strains (except for strain MC2013,which does not have an eno gene) and cluster 2 strains FC2001 and WCD2001 contain boththe N- and C-terminal domains and are indicated to be full length (432 amino acids [aa]).However, cluster 2 strains AC2005 and VCD2006 and cluster 1 strain AE2015 showedtruncated N-terminal domains, while cluster 2 strain XPD2002 had two predicted enolases.The first of these possessed truncated N- and C-terminal domains, and the second con-tained only the C-terminal Pfam domain with a predicted protein size of 129 aa. Thetruncated eno genes in cluster 2 strains AC2005 and VCD2006 and cluster 1 strain AE2015explain the lack of eno PCR products from these strains. The absence of an eno PCR productfor B. fibrisolvens NC3005 and cluster 2 strain WCD2001 is explained by an altered eno PCRprimer site at the 5= and 3= ends of these genes (Fig. S5B). The positive eno PCR result forXPD2002 is due to the eno primer site remaining intact in the shortened gene (129 aa).

The Butyrivibrio eno-negative strains and, possibly, the strains containing truncatedeno genes must use an alternative pathway for hexose metabolism. The methylglyoxalshunt is the most likely alternative pathway, which is mediated by the enzymes

FIG 5 Heat map of normalized relative abundances for CAZyme families determined for the Butyrivibrio and Pseudobutyrivibrio genomes. The relativenormalized inferred CAZy gene family abundances per genome (Z-score) are shown using a heat color scheme (red to green), indicating low to high relativeabundance. The Bray-Curtis distances (119) of compositional dissimilarity and hierarchical cluster analysis using Ward’s method (120) were used to calculatenormalized abundance. Genome names are colored to represent Butyrivibrio cluster 3 in green, Butyrivibrio cluster 2 in red, Butyrivibrio cluster 1 in blue, andPseudobutyrivibrio in purple.

Palevich et al. Applied and Environmental Microbiology

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 8

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from

fructose-1,6-bisphophate aldolase (Fbp), methylglyoxal synthase (MgsA), glyoxylase(GloA/B), S-lactoylglutathione hydrolase, and D-2-hydroxyacid dehydrogenase (LdhD),converting D-fructose-1,6-bisphophate to pyruvate (Fig. 6). The genes encoding thesekey methylglyoxal shunt enzymes were compared between the genomes of theButyrivibrio eno-positive and eno-negative strains (Fig. 6; Data Set S2). Analysis of thefermentation end products identified the production of lactate in Butyrivibrio strains(Data Set S1), which was exclusively L-lactate. Of particular interest was eno-negativeButyrivibrio sp. LC3010, which produced substantial amounts of L-lactate as a fermen-tation product but which also contained an incomplete set of methylglyoxal shuntpathway genes, in particular, glyoxalase I or lactoglutathione lyase (gloA) (Fig. 6; DataSet S2). Butyrivibrio sp. LC3010 and other such Butyrivibrio strains may thus usealternative enzymes to the methylglyoxal shunt that are yet to be characterized. Thegenes encoding lactate dehydrogenase (ldh) have been identified and compared in theButyrivibrio draft genome sequences (Fig. 6), in which the ldh gene encoding L-lactatedehydrogenase plays a key role in the production of L-lactate from pyruvate.

An alternative explanation for the lack of eno genes is that Butyrivibrio species may bespecialized pectin fermenters in the rumen. Pectin breakdown releases galacturonates andglucuronates, which are metabolized via 2-keto-3-deoxygluconate (KDG) rather than via theEMP pathway (Fig. 6). KDG is then converted to 2-keto-3-deoxygluconate phosphate(KDGP) by 2-dehydro-3-deoxygluconokinase and is then converted to pyruvate and

FIG 6 Comparisons of gene presence/absence for enzymes involved in the carbohydrate metabolic pathways in Butyrivibrio leading to the formation ofbutyrate, formate, acetate, and lactate. All metabolic pathways were compiled using information from the MetaCyc (121) and KEGG (122) databases. Thepresence or absence of genes encoding particular enzymes within genomes is indicated by full or empty cells, respectively in the panels. The order of genomesin the panels, from left to right, are as follows: row 1, Butyrivibrio cluster 3 (green) strains AB2020, FE2007, MD2001, ND3005, WTE3004, YRB2005, MC2013, andNC3005; row 2, Butyrivibrio cluster 2 (red) strains AE3006, MB2005, WCD3002, AD3002, WCD2001, VCD2006, AE3004, LC3010, WCE2006, AC2005, FC2001,XPD2002, and NC2002; row 3, Butyrivibrio cluster 1 (blue) strains NK4A153, AE2005, AE3003, LB2008, MB2003, AE2015, P6B7, B316T, FD2007, VCB2006, XBB1001,AE3009, MC2021, XPD2006, VCB2001, FCS006, NC2007, FCS014, and AE2032; and row 4, Pseudobutyrivibrio (purple) strains MA3014, HUN009, CF1b, AD2017,LB2011, and MD2005. The enolase-catalyzed reaction is shown in red, as the gene was absent from a number of Butyrivibrio strains. Color schemes for themetabolism pathways are as follows: the formation of formate in blue, acetate in green, butyrate in purple, L-lactate in red, and D-lactate by the proposedmethylglyoxal shunt in orange (69). Abbreviations: DHAP, dihydroxyacetone phosphate; DKI, 5-keto-4-deoxyuronate; DKII, 2,5-diketo3-deoxygluconate; KDG,2-keto-3-deoxygluconate; KDGP, 2-keto-3-deoxy-gluconate phosphate. Abbreviations for sugar transport systems are as follows: ABC, ATP binding cassette; MFS,major facilitator superfamily.

Polysaccharide-Degrading Capabilities of Butyrivibrio Applied and Environmental Microbiology

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 9

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from

glyceraldehyde-3-phosphate (GAP) by 2-keto-3-deoxygluconate 6-phosphate aldolase (40–42). All Butyrivibrio genomes encode the enzymes required to convert glucuronate throughto pyruvate (Fig. 6). Most strains also encode all the enzymes for galacturonate fermenta-tion through to pyruvate. The only step that is missing in cluster 3 Butyrivibrio sp. strainsMC2013 and NC3005 and cluster 2 strains AD3002, WCD2001, LC3010, WCE2006, andNC2002 is the tagaturonate conversion to altronate via the altronate/tagaturonate oxi-doreductase enzyme. This pathway is not known to generate ATP via electron transportphosphorylation (ETP), but it generates pyruvate, which potentially can lead to ATPproduction via the pathways described in Fig. 6.

DISCUSSION

Members of the genus Butyrivibrio are a major component of the ruminal microfloraand have been isolated from the gastrointestinal tracts and feces of various ruminants,monogastric animals, and humans (13–16, 21–25, 43, 44). Until 1996, all Butyrivibrio andPseudobutyrivibrio strains were assigned to a single species, Butyrivibrio fibrisolvens, dueto their phenotypic and metabolic similarities (8). Rumen Butyrivibrio and Pseudobutyr-ivibrio strains are currently divided into six species, represented by B. fibrisolvens, B.hungatei, B. proteoclasticus and B. crossotus and the Pseudobutyrivibrio species P.xylanivorans and P. ruminis. All of these species belong to the genetically diverseLachnospiraceae family, within the order Clostridiales (26). Based on 16S rRNA genesequences, the Butyrivibrio strains grouped into three clusters, designated Butyrivibriocluster 1, containing B. proteoclasticus (B316T), B. hungatei (JK615T), and 10 otherButyrivibrio strains; cluster 2, containing 12 Butyrivibrio strains; and cluster 3, containing8 strains, including the B. fibrisolvens type strain (D1T). Clusters 2 and 3 are eachphylogenetically cohesive, while the larger cluster, cluster 1, appears to contain acontinuum of related organisms containing at least two species. The FDG analyses gavephylogenetic associations consistent with the 16S rRNA gene sequence data, while theANI and AF identities of whole-genome nucleotide sequences between the proposedclusters of rumen Butyrivibrio strains varied considerably.

It is apparent that while the Butyrivibrio and Pseudobutyrivibrio genomes share about600 core genes, they also carry unique selections of genes drawn from the species’accessory genomes. Compared to Prevotella species from different sites in humans,Butyrivibrio species have a large number of orthologous gene families (45). Recent workhas shown that genes are gained and lost through the combined actions of gene loss,gene gain via lateral transfer, and gene duplication at higher rates in organisms on thetips of the phylogenetic tree (46, 47). Examples of Butyrivibrio species that may exhibitsuch plasticity at the genome level include Butyrivibrio cluster 1 strains B. proteoclasticusP6B7 and Butyrivibrio sp. strains LB2008 and AE2032, cluster 2 strain Butyrivibrio sp.strain NC2002, and cluster 3 strains Butyrivibrio sp. MC2013 and NC3005. This patterncould fit the scenario that among Butyrivibrio species accessory genes are transientlyadvantageous in only a small subset of strains.

The collective genome complement (29,105 genes) and the core genome (602genes) of all rumen Butyrivibrio and Pseudobutyrivibrio strains reflect a large reservoir ofgenetic diversity within this group (Fig. 2). The strict core genome represents 2% of thecollective genome and represents the proposed minimum set of genes that allow thesurvival of Butyrivibrio species in the rumen, including genes encoding protein pro-cessing, folding and secretion (predominantly translation, including ribosome function,maturation, modification, protein turnover, and RNA degradation), cellular processes(cell division and transport), and energy and metabolism (lipid metabolism and bio-synthesis of nucleotides and cofactors) and numerous poorly characterized genes(conserved hypothetical proteins, etc.). The size of the collective Butyrivibrio genomeincreased with the number of genomes analyzed due to unique strain-specific genesand reflects the ability of this group of organisms to occupy different niches within therumen environment.

Early studies reported the presence of large extrachromosomal elements in anumber of Butyrivibrio strains (48), and recently, megaplasmids and chromids were

Palevich et al. Applied and Environmental Microbiology

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 10

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from

described for B. proteoclasticus B316T and B. hungatei MB2003 (49–53). The presentstudy has confirmed that extrachromosomal chromids and megaplasmids are commonin Butyrivibrio species and possibly improve their competitiveness by increasing thenumber of genes carried in the bacterial genome through gene dosing effects andallow for faster genome replication and a higher growth rate of the bacterial cell (54).A rumen plasmidome study identified phylogenetic associations of various plasmidgenes across taxonomic levels up to the phylum level, emphasizing the essentialevolutionary and cooperative roles between plasmids and their host bacteria (55). Dueto the incidence of related plasmids in phylogenetically distant bacteria, coupled withthe ability to be horizontally transferred by conjugation, plasmids likely play a role asa channel for the horizontal exchange of genomic material, conveying advantageousfunctions between the rumen microbes. The types of traits that are transferred byplasmids include those implicated in amino acid, protein, and carbohydrate metabo-lism, which are essential in the rumen ecosystem, and make the metabolic burden ofsustaining the plasmid worthwhile for the host (56). The transfer of genes encodingdegradative systems (57), exopolysaccharide production (58), bacteriocin production(59, 60), and resistance to antibiotics (61, 62) may also provide a competitive advantagewithin the rumen microbial ecosystem. In Butyrivibrio, it is possible that extrachromo-somal elements serve as vehicles for the exchange of genomic information betweendifferent strains and species and potentially to other genera, such as Pseudobutyrivibrio.

Early characterizations of Butyrivibrio strains indicated that some had the ability todegrade cellulose and that most strains were able to digest xylan and pectin substrates(8, 10). Later work concluded that glucose, cellobiose, maltose, and esculin wereuniversally used substrates of Butyrivibrio strains (30). Rumen Butyrivibrio and Pseudobu-tyrivibrio strains are reported to use a wide range of soluble and some insolublesubstrates and characteristically ferment carbohydrates to butyrate, formate, lactate,and acetate (26, 30). The findings from the present study show the absence of growthon cellulose by all 30 Butyrivibrio strains as well as P. xylanivorans MA3014. The volatilefatty acid (VFA) production data support the notion that rumen Butyrivibrio strains aremetabolically versatile and can utilize a wide range of insoluble substrates but are notcellulose-degrading bacteria. The substrate utilization patterns of the Butyrivibrio spe-cies generally followed the groupings defined by the phylogenetic and genomicanalyses. For example, the B. fibrisolvens MD2001, AB2020, FE2007, ND3005, YRB2005,and WTE3004 strains displayed similar growth patterns, in particular, their ability toutilize pectin, xylan, and xylose, whereas cluster 3 strains MC2013 and NC3005 ex-pressed different growth patterns that coincided with their phylogenetic and genomicdivergence within this cluster (Fig. 1). All cluster 2 strains could utilize xylose, andstrains AE3004, XPD2002, and VCD2006 displayed similar growth patterns on allinsoluble substrates analyzed. Cluster 1 strains B316T, VCB2006, XBB1001, and AE3009,which grouped closely at the genome and phylogenetic levels, also displayed similargrowth patterns across all insoluble substrates (except cellulose) and monosaccharidesanalyzed.

The comparative genome analyses identified considerable variation in the con-servation of orthologous gene families both between and within the rumen Butyr-ivibrio strains. This suggests a degree of specialization within these bacteria, inaddition to the presence of a set of genes required for polysaccharide degradationin the rumen. The analyses of the Butyrivibrio and Pseudobutyrivibrio CAZymesinvolved in the breakdown of complex carbohydrates showed variation in theirdistribution and abundance. The abundance of GH, PL, and CE domain-containingCAZymes encoded within the genomes of the cluster 1 and 2 strains suggests thatthey are specialist degraders of xylan and pectin (see Data Set S2 in the supple-mental material). This suggests that the members of cluster 1 play an important rolein polysaccharide degradation in the rumen and are significant energy suppliers toruminants through the production of VFAs. The growth experiments and compar-ative glycobiome analyses have shown that strains belonging to this species group

Polysaccharide-Degrading Capabilities of Butyrivibrio Applied and Environmental Microbiology

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 11

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from

are diverse in both the substrates that they can utilize and the set of CAZymes thatthey encode and may occupy similar niches in the rumen.

A previous analysis of the B. proteoclasticus B316T genome described that two-thirdsof its CAZymes involved in polysaccharide breakdown were predicted to be intracellularand that only a few CBMs were present (49), and this was also true for the strains in thisstudy. These observations indicate that the ability to degrade plant fiber and utilize thereleased carbohydrates for growth are important features defining the differencesbetween Butyrivibrio strains and may reflect how these strains occupy different nicheswithin the rumen in order to coexist. Investigations of the xylan and pectin utilizationabilities of B. hungatei MB2003 and B. proteoclasticus B316T in coculture support thisview (63). In monocultures, B316T was able to grow well on xylan and pectin, whileMB2003 was unable to utilize either of these insoluble substrates to support significantgrowth. Cocultures of B316T grown with MB2003 revealed that MB2003 showed growthalmost equivalent to that of B316T when either xylan or pectin was supplied as thesubstrate. The effect of coculture on the transcriptomes of B316T and MB2003 wasassessed, where B316T transcription was largely unaffected by the presence of MB2003.However, MB2003 expressed a wide range of genes encoding proteins for carbohydratedegradation, central metabolism, oligosaccharide transport, and substrate assimilation,in order to compete with B316T for the released sugars. These results suggest thatB316T has a role as an initiator of the primary solubilization of xylan and pectin, whileMB2003 competes effectively for the released soluble sugars to enable its growth andmaintenance in the rumen.

Butyrivibrio and Pseudobutyrivibrio have previously been grouped into lactate-producing and lactate-nonproducing strains (30). Also, the diversity in fermentationproducts observed within each Butyrivibrio cluster suggests differences in the metabolicpathways of each strain and further highlights the importance and adaptive role ofButyrivibrio in the digestion of fibrous constituents of animal feed. The present studyhas shown an ability of Butyrivibrio as a genus to utilize a wide variety of substrates,especially xylan and pectin, suggesting that these microorganisms play an importantrole in hemicellulose and pectin degradation in the rumen. In particular, the productionof large amounts of VFAs by B. proteoclasticus B316T and Butyrivibrio sp. strain AE3009on a range of insoluble substrates indicates the ability of certain Butyrivibrio to switchsubstrate utilization from the simple cellobiose substrate to insoluble polysaccharidesin monoculture. Therefore, it is hypothesized that certain Butyrivibrio strains are unableto initiate significant degradation of the insoluble polysaccharides alone and rely onmore specialized Butyrivibrio species to initiate the degradation process, resulting in therelease of soluble sugars, for which they compete.

Given the important role of rumen Butyrivibrio strains in plant fiber degradation (6, 7, 63),genome sequence information was used to analyze their collective and individualpolysaccharide-degrading potential. There are examples of contrasting differences in the abun-dance of some GH families in Butyrivibrio versus Pseudobutyrivibrio. For instance, Butyrivibriostrains have many members of GHs 28, 30, 38, 65, 67, 88, 105, 112, and 129 (Data Set S2), withpredicted activities such as polygalacturonases, �-xylosidases, �-mannosidases, �-trehalases,�-glucuronidases, �-glucuronyl hydrolases, rhamnogalacturonyl hydrolases, lacto-N-biose phos-phorylases, and �-N-acetylgalactosaminidases, respectively (Data Set S2). In contrast, these werecompletely absent from the six Pseudobutyrivibrio strains. Analysis of the CAZy profiles of therumen Butyrivibrio species presented here suggests that, due to their extensive repertoire of GHdomain-containing CAZymes, Butyrivibrio cluster 1 and 3 strains rather than Pseudobutyrivibriostrains are the predominant degraders of xylan and pectin.

The carbon source utilization and VFA data combined with genome similarity andCAZyme analyses showed that Butyrivibrio utilization of polysaccharides and the abilityof the Butyrivibrio strains to assimilate the degradation products of these polysaccha-rides were variable, with clear differences being identified between species groupsbased on their initial phylogenetic placements. For the production of butyrate and H2

from glucose, rumen Butyrivibrio genomes possess a pyruvate:ferredoxin oxidoreduc-tase gene (nifJ) required for pyruvate conversion to acetyl coenzyme A (CoA), as well as

Palevich et al. Applied and Environmental Microbiology

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 12

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from

a butyryl-CoA dehydrogenase/electron-transferring flavoprotein (encoded by bcd-etfAB) to generate ATP by classic substrate-level phosphorylation (SLP). In addition, analternative pathway exists where formate is predicted to be the end product andinvolves the decarboxylation of acetyl-CoA by a pyruvate formate lyase (encoded bypflB) instead of NifJ. It has been proposed that Ech and Rnf work in concert with NifJand the Bcd-Etf complex to drive ATP synthesis by ETP during glucose fermentation tobutyrate (64–66). Interestingly, the vast majority of anaerobic prokaryotes appear topossess either an Ech or an Rnf protein but not both (67, 68). However, a recent analysisof rumen prokaryotic genomes identified rumen Butyrivibrio species to be a rare groupof bacteria that possess genes for both Ech and Rnf. These findings warrant furtherbiochemical investigation to determine the activity of Ech and Rnf in Butyrivibrio.

In ruminal anaerobes, hexoses are usually fermented via the Embden-Meyerhof-Parnas (EMP) glycolytic pathway to pyruvate and from pyruvate to a variety of endproducts, depending on the organism. In Butyrivibrio species, these end products areprincipally formate and butyrate, with a small amount of acetate, while some strains arealso capable of producing lactate. The rumen Butyrivibrio pathways for butyrate pro-duction presume the possession of a complete EMP glycolytic pathway (Fig. 6). Enolase(encoded by eno; EC 4.2.1.11) converts 2-phospho-D-glycerate to phosphoenolpyruvatein the second-to-last step of the EMP pathway. In the proposed alternative methylg-lyoxal shunt pathway (49), the dihydroxyacetone phosphate (DHAP) is transformed topyruvate via methylglyoxal and D-lactate dehydrogenase, encoded by ldhD (69). Therumen Butyrivibrio genomes presented here have the same set of genes previouslyreported for MB2003 (52, 63) and B316T (49) for the production of butyrate, formate,acetate, and lactate, and these genes appear to be common features among theserumen organisms.

Butyrivibrio and Pseudobutyrivibrio species form a significant group of rumen bac-teria that play an important role in the carbon flow within the rumen by initiating thebreakdown of lignocellulose and metabolizing the by-products to short-chain fattyacids and fermentation end products. Butyrivibrio and Pseudobutyrivibrio strains encodea large and diverse spectrum of degradative CAZymes and binding proteins. In total,4,421 GHs, 1,283 CEs, 110 PLs, 3,605 GTs, and 1,706 CBMs with predicted activitiesinvolved in the depolymerization and utilization of the insoluble plant polysaccharides,such as xylan and pectin, were identified. The different Butyrivibrio species were foundto possess similar CAZyme repertoires, but with variations in the absolute number ofgenes within each CAZy category. This apparent functional redundancy encoded byclosely related strains was also observed with examination of both 16S rRNA markergene and genome sequence-based species group demarcations. Herein, a significantexample of gene loss was highlighted by the absence of an identifiable enolase, a keyenzyme that drives the penultimate step of glycolysis, in the majority of Butyrivibriostrains. The comparative genome analyses provide further evidence for the need toinclude genome sequencing as a prerequisite for the description of new species ofbacterial isolates. Together with previous gene expression data, our findings suggestthat members of the genera Butyrivibrio and Pseudobutyrivibrio occupy similar nichesbut apply different degradation strategies within the rumen.

MATERIALS AND METHODSCultures used in this study and growth conditions. The full list of cultures, their provenance, and

the phenotypic characteristics for a selection of the Butyrivibrio isolates used in the project are shown inData Set S1 in the supplemental material. The bacterial cultures used in this study were grown aspreviously described (37, 70). New Zealand Butyrivibrio and Pseudobutyrivibrio cultures from the HungateCollection are available from the AgResearch culture collection (7).

Bacterial genomic DNA isolation and identification. Genomic DNA was extracted using a QiagenGenomic-tip kit following the manufacturer’s instructions for the 500/G size extraction. Extracted DNAwas stored at �80°C until required. Purified DNA was subject to partial 16S rRNA gene sequencing toconfirm strain identity, before being shipped to the U.S. Department of Energy (DOE) Joint GenomeInstitute (JGI), USA, for sequencing.

Phylogenetic analysis of full-length 16S rRNA gene sequences. To determine the phylogeneticrelationships of the Butyrivibrio isolates shown in Fig. S1, the extracted DNA was PCR amplified using theprimer pair fD1 (5=-GAGTTTGATCMTGGCTCAG-3=) and rD1 (5=-AAGGAGGTGATCCARCCG-3=) to amplify

Polysaccharide-Degrading Capabilities of Butyrivibrio Applied and Environmental Microbiology

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 13

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from

the V1-V3 region of the 16S rRNA gene. The PCR cycling conditions used were 94°C for 2 min, followedby 35 cycles of 94°C for 30 s, 56°C for 30 s, and 72°C for 2 min and a final extension time of 10 min at 72°C.The full-length 16S rRNA marker gene sequences (�1,400 bp) obtained by PCR amplification were Sangersequenced using eight primers: fD1, rD1, 1492r, 1382r, 1100r, 806r, 514f, and 514r (71–75). The Staden(76) and the Geneious (77) software packages were used for trimming and aligning the forward andreverse sequences. The trimmed sequences were manually assessed and compared against the NationalCenter for Biotechnology Information (NCBI) nonredundant nucleotide database (http://blast.ncbi.nlm.nih.gov) using the MegaBLAST algorithm (78) available on NCBI’s Basic Local Alignment Search Tool(BLAST) web interface (79) and the Ribosomal Database Project (RDP) (80). Identity was determined bydatabase matches with E values of 0, identity of 99 to 100%, and 100% coverage. The global alignmentof the nearly full-length 16S rRNA gene nucleotide sequences was performed using the ClustalWprogram (81), and phylogenetic analyses were performed using Molecular Evolutionary Genetics Anal-ysis, version 6.0 (MEGA6), software (82). Phylogenetic trees were constructed using the neighbor-joiningmethod (83), with distances being calculated using the Kimura 2-parameter method (84) and pairwisedeletions of gaps. The phylogeny based on the 16S rRNA gene sequence was inferred using themaximum likelihood (ML) method (85). Bootstrap analysis with 10,000 replicates was used to assess thestatistical strength of the branch positions (84). The 16S rRNA gene sequence from Methanobrevibacterruminantium M1 (GenBank accession number CP001719) was used as an outgroup for the tree.

Cell motility and flagellar biosynthesis operons. To examine the motility of the Butyrivibrio strains,a motility agar stab test was carried out using 5 ml of 0.3% (wt/vol) agar in RM02 medium with cellobioseas the carbon source in a 10-ml Hungate tube (37). Freshly grown Butyrivibrio cultures were stabinoculated into the agar using a straightened inoculating loop, and the cultures were incubated at 37°Cfor 24 to 48 h. Nonmotile strains displayed visible growth confined to the area immediately surroundingthe initial inoculation stab. Motile bacterial cells migrated through the agar to produce a diffuse orcloudy growth pattern, as evidenced by turbidity located a distance away from the inoculation stab (86).The strains used as negative controls included B. hungatei MB2003 (52), B. proteoclasticus B316T (87),Prevotella ruminicola 23 (88), and Streptococcus bovis 2B (89), while P. xylanivorans MA3014 (37, 50) wasused as a positive control. The variable regions of the flagellum biosynthesis operon were compared inButyrivibrio strains and correlated with their motility in order to determine the relationship between thegenotype and phenotype (90). Annotations from the Integrated Microbial Genomes with Metagenomes(IMG/M) system (91) were used to validate the functionality of the genes within the flagellum biosyn-thesis operons. The gene sequences were manually assessed using BLASTn from the BLAST� package(92).

Carbohydrate source utilization and fermentation end product analysis. Bacterial carbon sourceutilization was tested on fresh Butyrivibrio cultures inoculated into Hungate tubes containing RM02 mediumbroth with the separate addition of each of the 32 carbon sources, including seven polysaccharides (glycogen,pectin, inulin, cellulose, dextrin, starch, and xylan) at a 0.5% (wt/vol) final concentration (37, 50). Cultures wereinoculated in triplicate and grown under anaerobic conditions overnight at 39°C. The optical density at 600nm (OD600) readings were measured initially after inoculation and after 24 h and 72 h of incubation. For thesoluble substrates, changes in the OD600 (ΔOD600) readings of 0.5 to 1.0 were scored as ��, changes of 0.2to 0.5 were scored as �, and changes of 0 to 0.2 were scored as �. Positive controls were individual strainsgrown in cultures containing D-glucose, D-cellobiose, D-xylose, and L-arabinose. Two types of negative controlswere included: inoculation controls without a carbon source added and uninoculated medium with thesubstrate. Uninoculated medium without the substrate was used as a blank. Growth on polysaccharidesubstrates was assessed by measurement of volatile fatty acid (VFA) production. VFA production wasdetermined from triplicate broth cultures grown overnight with cellobiose as the substrate and analyzed forformate, acetate, propionate, n-butyrate, isovalerate, and lactate on an HP 6890 series gas chromatograph(Hewlett-Packard) with 2-ethylbutyric acid (Sigma-Aldrich) as the internal standard. To derivatize formic, lactic,and succinic acids, samples were mixed with the HCl ACS reagent (Sigma-Aldrich) and diethyl ether, with theaddition of N-methyl-N-t-butyldimethylsilyltri-fluoroacetamide (MTBSTFA) (Sigma-Aldrich) (93). A D-lactic acidassay kit and L-lactic acid assay kit (Megazyme Inc., Bray, Ireland) were used for measurements of D- andL-lactate concentrations, respectively. All samples were diluted to yield a lactic acid concentration of 0.03 to0.30 g/liter, the linear range of the assay. The microplate assay procedure was performed according to themanufacturer’s instructions with a 224-�l reaction volume.

Screening for enolase genes. For comparisons of enolase gene (eno)-positive versus enolasegene-negative Butyrivibrio strains, the eno genes were identified and annotated based on the IntegratedMicrobial Genomes (IMG) system of identification of enolase Pfam (C-terminal Pfam00113 and N-terminalPfam03952), COG (COG0148), KOG (KOG2670), and KO (KO1689) domains. In addition, the Metastatsprogram (94) was employed in conjunction with contrasting upper and lower quartile or percentile genecounts, in order to identify additional functions with a pattern of preservation/loss similar to that of theglycolytic enolase gene (7). The eno gene Pfam domains were compared for enolase-positive Butyrivibriostrains, and the respective amino acid sequences of the enolase proteins were compared using amaximum likelihood (ML) alignment analysis. Genomic DNAs from a selection of Butyrivibrio strains wereextracted as described above and screened for the presence of enolase genes by PCR amplification usingthe forward primer 5=-AATGGACCTAYGCAGATGC-3= and reverse primer 5=-ATCTGGTTRAGCTTWATAAG-3= (49). The PCR cycling conditions used were 94°C for 2 min, followed by 35 cycles of 94°C for 30 s,50°C for 30 s, and 72°C for 2 min and a final extension time of 10 min at 72°C. The PCR products wereanalyzed by agarose gel electrophoresis, and the concentrations were determined with a NanoDropspectrophotometer and a Qubit double-stranded DNA BR assay kit (Thermo Fisher Scientific Inc.). Toinvestigate the possibility that the enolase primers did not detect all enolase genes, the enolase primer

Palevich et al. Applied and Environmental Microbiology

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 14

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from

sequences were aligned against the draft genomes of the Butyrivibrio strains. A strong alignment of boththe forward and the reverse degenerative primers was achieved for all enolase-positive Butyrivibriostrains that screened positive for presence of eno.

Sequence assembly and annotation. All Hungate genomes were sequenced at the DOE JointGenome Institute (JGI) using the Illumina technology (95) or Pacific Biosciences (PacBio) RS technology(96). For all genomes, we either constructed or sequenced an Illumina short-insert paired-end library withan average insert size of 270 bp or a PacBio SMRTbell library. Genomes were assembled using the Velvet(97), ALLPATHS (98) or Hierarchical Genome Assembly Process (HGAP) (99) assembly methods. Genomeswere annotated by the DOE JGI genome annotation pipeline (100, 101). Briefly, protein-coding genes(coding sequence [CDSs]) were identified using the Prodigal program (102), followed by a round ofautomated and manual curation using the JGI GenePrimp pipeline (103). Functional annotation andadditional analyses were performed within the Integrated Microbial Genomes Expert Review (IMG-ER)platform (91).

Comparative analysis of the genome data sets. (i) CAZyme annotation. The putative proteomes ofthe 40 Butyrivibrio and 6 Pseudobutyrivibrio data sets were subjected to automated annotation and assignment toCAZymes using the dbCAN resource CAZy family-specific hidden Markov models (HMMs) (104). An E value of�1e�3 for CAZymes based on family-specific HMMs was used as the cutoff for alignments shorter than 80 aminoacids, while an E value of �1e�5 was used for alignments longer than 80 amino acids. These cutoff settings enabledshort but significant CBM matches to be maintained. All dbCAN hits were clustered at a 100% sequence identitythreshold using the CD-HIT Illumina algorithm to remove duplicates (105). All descriptions and classifications werecompiled from CAZy (106), and the modular architectures of CAZymes and predicted proteins with multimodularCAZyme organizations in the genome data sets were determined by searching each query protein against thePfam and Protein Data Bank (PDB) databases (107, 108).

(ii) ANI and AF computation. ANI and the fraction of orthologous genes (AF) were used as complementarymeasures of genetic relatedness based on the gene content between the 40 Butyrivibrio and 6 Pseudobutyrivibriogenomes. ANI is a measure of nucleotide-level genomic similarity between the coding regions of two genomes,determined using a custom Perl script and the high-performance similarity search tools NSimScan and PSimScan(109). Each genome sequence served as a reference genome, and the resulting ANI values were averaged. The codeto perform genomic ANI and AF computation is available at https://ani.jgi.doe.gov/html/download.php. The AFand ANI were calculated for the 40 Butyrivibrio and 6 Pseudobutyrivibrio genomes to determine speciescutoffs. In order to identify species ANI and AF that determine whether the genomes in a pair belong tothe same species, only the subset of high-quality genome pairs was utilized. An ANI cutoff of �96.5 andan AF cutoff of �0.6 were used to define species.

(iii) FGD. The functional genome distribution (FGD) is a tool for comparative microbial genomics analysis andinterpretation of the genetic diversity of bacteria (110). FGD investigates the overall similarity levels betweenmicrobial genomes, based on the amino acid sequences of their predicted ORFeomes, which correspond to thecoding sequences (CDSs) of the genes (open reading frames [ORFs]) in a genome, and ultimately defines thedegree of similarity of the genomes. All Hungate Butyrivibrio and Pseudobutyrivibrio genomes were downloaded inFASTA format from the IMG genome database (111), concatenated using a universal spacer-stop-spacer sequence,and automatically annotated using the GAMOLA2 software package (112). The in-house closed genomes of B.proteoclasticus B316T (49) and B. hungatei MB2003 (52) and the draft genome of P. xylanivorans MA3014 weremanually annotated using GAMOLA2 (112). The predicted ORFeomes of all genomes were subjected to an FGDanalysis, and the resulting distance matrix was imported into MEGA6 (82). The functional genome distribution wasvisualized using the unweighted pair group method with arithmetic mean (UPGMA) method (113, 114).

(iv) Genome alignment. A MUMmer (version 3) system was used to compare the alignment ofcontigs across a reference genome in the form of a dot plot diagram (115). Synteny plots were generatedusing the mummerplot utility. The plots reveal regions of exact matches between the pair of genomescompared and thus are an indicator of the conservation between the two genomes. The Gsview program(116) was used to visualize the generated MUMmer plot.

(v) Amino acid and codon usage analyses. A comparison of the amino acid and codon usage betweenthe 40 Butyrivibrio and 6 Pseudobutyrivibrio genomes was performed using the CMG-biotools package (117) underthe default parameters. Amino acid and codon usages were calculated using BioPerl modules, which calculate eachamino acid or codon count as a fraction of the total count of amino acids or codons. The percent codon and aminoacid usage was plotted in two-dimensional heat maps using gplots in R (118), reordering the organisms and theamino acids/codon to show the shortest distance between them. Dendrograms were used to visualize thedifference in usage between different strains.

(vi) Determination of the core and pan-genomes. The genes representative of the Butyrivibrio andPseudobutyrivibrio core and pan-genomes were determined by performing a BLAST-based analysis usingthe CMG-biotools package (117) with default parameters. If two proteins within a genome met thedesignated cutoff, they were clustered into one protein family. Protein families were extended viasingle-linkage clustering. If a protein family included proteins from all genomes in the comparison, thefamily was designated a core protein family. Subset genes, such as species group shared and uniquesubsets of genes within individual genomes, were identified by clustering the results from the core andpan-genome calculations.

Data availability. The data sets supporting the conclusions of this article are available through theIMG portal (https://img.jgi.doe.gov/). Additionally, a dedicated portal to download all genomes se-quenced as part of the Hungate1000 project (7) is provided at https://genome.jgi.doe.gov/portal/pages/dynamicOrganismDownload.jsf?organism�HungateCollection.

Polysaccharide-Degrading Capabilities of Butyrivibrio Applied and Environmental Microbiology

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 15

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from

SUPPLEMENTAL MATERIALSupplemental material is available online only.SUPPLEMENTAL FILE 1, PDF file, 1.3 MB.SUPPLEMENTAL FILE 2, XLSX file, 0.2 MB.SUPPLEMENTAL FILE 3, XLSX file, 0.1 MB.

ACKNOWLEDGMENTSThis work was supported by the New Zealand Ministry of Business, Innovation and Employ-

ment New Economy Research Fund program Accessing the uncultured rumen microbiome(contract number C10X0803). The Hungate1000 project was funded by the New ZealandGovernment in support of the Livestock Research Group of the Global Research Alliance onAgricultural Greenhouse Gases (http://www.globalresearchalliance.org). The genome se-quencing and analysis component of the project was supported by the U.S. Depart-ment of Energy Joint Genome Institute (JGI) through its Community Science Program(CSP 612) under contract number DE-AC02-05CH11231 and used the resources of theNational Energy Research Scientific Computing Center, which is supported by the Officeof Science of the U.S. Department of Energy.

We acknowledge and thank all the JGI staff that contributed to this project andSarah Lewis for assistance with the fermentation end product analysis.

REFERENCES1. Godfray HCJ, Beddington JR, Crute IR, Haddad L, Lawrence D, Muir JF,

Pretty J, Robinson S, Thomas SM, Toulmin C. 2010. Food security: thechallenge of feeding 9 billion people. Science 327:812– 818. https://doi.org/10.1126/science.1185383.

2. Eisler MC, Lee MRF, Tarlton JF, Martin GB, Beddington J, Dungait JAJ,Greathead H, Liu J, Mathew S, Miller H, Misselbrook T, Murray P, VinodVK, Van Saun R, Winter M. 2014. Agriculture: steps to sustainablelivestock. Nature 507:32–34. https://doi.org/10.1038/507032a.

3. Hess M, Sczyrba A, Egan R, Kim T-W, Chokhawala H, Schroth G, Luo S,Clark DS, Chen F, Zhang T, Mackie RI, Pennacchio LA, Tringe SG, Visel A,Woyke T, Wang Z, Rubin EM. 2011. Metagenomic discovery of biomass-degrading genes and genomes from cow rumen. Science 331:463– 467.https://doi.org/10.1126/science.1200387.

4. Koike S, Yoshitani S, Kobayashi Y, Tanaka K. 2003. Phylogenetic analysisof fiber-associated rumen bacterial community and PCR detection ofuncultured bacteria. FEMS Microbiol Lett 229:23–30. https://doi.org/10.1016/S0378-1097(03)00760-2.

5. McAllister TA, Bae HD, Jones GA, Cheng KJ. 1994. Microbial attachmentand feed digestion in the rumen. J Anim Sci 72:3004 –3018. https://doi.org/10.2527/1994.72113004x.

6. Henderson G, Cox F, Ganesh S, Jonker A, Young W, Global RumenCensus Collaborators, Janssen PH. 2015. Rumen microbial communitycomposition varies with diet and host, but a core microbiome is foundacross a wide geographical range. Sci Rep 5:14567. https://doi.org/10.1038/srep14567.

7. Seshadri R, Leahy SC, Attwood GT, Teh KH, Lambie SC, Cookson AL,Eloe-Fadrosh EA, Pavlopoulos GA, Hadjithomas M, Varghese NJ, Paez-Espino D, Hungate1000 Project Collaborators, Perry R, Henderson G,Creevey CJ, Terrapon N, Lapebie P, Drula E, Lombard V, Rubin E,Kyrpides NC, Henrissat B, Woyke T, Ivanova NN, Kelly WJ. 2018. Culti-vation and sequencing of rumen microbiome members from the Hun-gate1000 Collection. Nat Biotechnol 36:359 –367. https://doi.org/10.1038/nbt.4110.

8. Bryant MP, Small N. 1956. The anaerobic monotrichous butyric acid-producing curved rod-shaped bacteria of the rumen. J Bacteriol 72:16 –21.

9. Van Gylswyk N, Hippe H, Rainey F. 1996. Pseudobutyrivibrio ruminis gen.nov., sp. nov., a butyrate-producing bacterium from the rumen thatclosely resembles Butyrivibrio fibrisolvens in phenotype. Int J Syst EvolMicrobiol 46:559 –563. https://doi.org/10.1099/00207713-46-2-559.

10. Hungate RE. 1966. The rumen and its microbes. Academic Press, NewYork, NY.

11. Diez-Gonzalez F, Bond DR, Jennings E, Russell JB. 1999. Alternativeschemes of butyrate production in Butyrivibrio fibrisolvens and their

relationship to acetate utilization, lactate production, and phylogeny.Arch Microbiol 171:324 –330. https://doi.org/10.1007/s002030050717.

12. Paillard D, McKain N, Rincon MT, Shingfield KJ, Givens DI, Wallace RJ.2007. Quantification of ruminal Clostridium proteoclasticum by real-timePCR using a molecular beacon approach. J Appl Microbiol 103:1251–1261. https://doi.org/10.1111/j.1365-2672.2007.03349.x.

13. Hespell RB, Cotta MA. 1995. Degradation and utilization by Butyrivibriofibrisolvens H17c of xylans with different chemical and physical prop-erties. Appl Environ Microbiol 61:3042–3050.

14. Bryant MP. 1986. Genus IV. Butyrivibrio, p 641– 643. In Sneath PHA, MairNS, Sharpe ME, Holt JG (ed), Bergey’s manual of systematic bacteriol-ogy, vol 2. Williams & Williams, Baltimore, MD.

15. Hespell RB. 1992. The genera Butyrivibrio, Lachnospira, and Roseburia, p2023–2033. In Balows A, Trüper HG, Dworkin M, Harder W, Schleifer K-H(ed), The prokaryotes: a handbook on the biology of bacteria: ecophysi-ology, isolation, identification, applications. Springer, New York, NY.

16. Miron J. 1991. The hydrolysis of Lucerne cell-wall monosaccharidecomponents by monocultures or pair combinations of defined ruminalbacteria. J Appl Bacteriol 70:245–252. https://doi.org/10.1111/j.1365-2672.1991.tb02932.x.

17. Orpin CG, Mathiesen SD, Greenwood Y, Blix AS. 1985. Seasonal changesin the ruminal microflora of the high-arctic Svalbard reindeer (Rangifertarandus platyrhynchus). Appl Environ Microbiol 50:144 –151.

18. Rumney CJ, Duncan SH, Henderson C, Stewart C. 1995. Isolation andcharacteristics of a wheatbran-degrading Butyrivibrio from human fae-ces. Lett Appl Microbiol 20:232–236. https://doi.org/10.1111/j.1472-765x.1995.tb00435.x.

19. Dehority B. 1966. Characterization of several bovine rumen bacteriaisolated with a xylan medium. J Bacteriol 91:1724 –1729.

20. Hespell RB. 1992. Fermentation of xylans by Butyrivibrio fibrisolvens andThermoanaerobacter strain B6A: utilization of uronic acids and xylano-lytic activities. Curr Microbiol 25:189 –195. https://doi.org/10.1007/BF01570718.

21. Miron J, Ben-Ghedalia D. 1993. Digestion of cell-wall monosaccharidesof ryegrass and alfalfa hays by the ruminal bacteria Fibrobacter succi-nogenes and Butyrivibrio fibrisolvens. Can J Microbiol 39:780 –786.https://doi.org/10.1139/m93-115.

22. Williams AG, Withers SE. 1992. Induction of xylan-degrading enzymesin Butyrivibrio fibrisolvens. Curr Microbiol 25:297–303. https://doi.org/10.1007/BF01575865.

23. Blackburn T, Hobson P. 1962. Further studies on the isolation of pro-teolytic bacteria from the sheep rumen. J Gen Microbiol 29:69 – 81.https://doi.org/10.1099/00221287-29-1-69.

24. McKain N, Shingfield KJ, Wallace RJ. 2010. Metabolism of conjugatedlinoleic acids and 18:1 fatty acids by ruminal bacteria: products and

Palevich et al. Applied and Environmental Microbiology

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 16

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from

mechanisms. Microbiology 156:579 –588. https://doi.org/10.1099/mic.0.036442-0.

25. Maia MR, Chaudhary LC, Figueres L, Wallace RJ. 2007. Metabolism ofpolyunsaturated fatty acids and their toxicity to the microflora of therumen. Antonie Van Leeuwenhoek 91:303–314. https://doi.org/10.1007/s10482-006-9118-2.

26. pMoon CD, Pacheco DM, Kelly WJ, Leahy SC, Li D, Kopecny J, Attwood GT.2008. Reclassification of Clostridium proteoclasticum as Butyrivibrio prote-oclasticus comb. nov., a butyrate-producing ruminal bacterium. Int J SystEvol Microbiol 58:2041–2045. https://doi.org/10.1099/ijs.0.65845-0.

27. Forster RJ, Gong J, Teather RM. 1997. Group-specific 16S rRNA hybrid-ization probes for determinative and community structure studies ofButyrivibrio fibrisolvens in the rumen. Appl Environ Microbiol 63:1256 –1260.

28. Willems A, Amat-Marco M, Collins MD. 1996. Phylogenetic analysis ofButyrivibrio strains reveals three distinct groups of species within theClostridium subphylum of the gram-positive bacteria. Int J Syst Bacte-riol 46:195–199. https://doi.org/10.1099/00207713-46-1-195.

29. Forster R, Teather R, Gong J, Deng S-J. 1996. 16S rDNA analysis ofButyrivibrio fibrisolvens: phylogenetic position and relation to butyrate-producing anaerobic bacteria from the rumen of white-tailed deer. LettAppl Microbiol 23:218 –222. https://doi.org/10.1111/j.1472-765x.1996.tb00069.x.

30. Kopecný J, Zorec M, Mrazek J, Kobayashi Y, Marinšek-Logar R. 2003.Butyrivibrio hungatei sp. nov. and Pseudobutyrivibrio xylanivorans sp.nov., butyrate-producing bacteria from the rumen. Int J Syst EvolMicrobiol 53:201–209. https://doi.org/10.1099/ijs.0.02345-0.

31. Dibbayawan T, Cox G, Cho KY, Dwarte DM. 1985. Cell wall and plasmamembrane architecture of Butyrivibrio spp. J Ultrastructure Res 90:286 –293. https://doi.org/10.1016/S0022-5320(85)80006-X.

32. Bryant MP, Robinson IM. 1962. Some nutritional characteristics of pre-dominant culturable ruminal bacteria. J Bacteriol 84:605– 614.

33. Roché C, Albertyn H, van Gylswyk NO, Kistner A. 1973. The growthresponse of cellulolytic acetate-utilizing and acetate-producing butyr-ivibrios to volatile fatty acids and other nutrients. J Gen Microbiol78:253–260. https://doi.org/10.1099/00221287-78-2-253.

34. Shane BS, Gouws L, Kistner A. 1969. Cellulolytic bacteria occurring inthe rumen of sheep conditioned to low-protein teff hay. J Gen Micro-biol 55:445– 457. https://doi.org/10.1099/00221287-55-3-445.

35. Hazlewood GP, Theodorou MK, Hutchings A, Jordan DJ, Galfre G. 1986.Preparation and characterization of monoclonal antibodies to a Butyr-ivibrio sp. and their potential use in the identification of rumen butyr-ivibrios, using an enzyme-linked immunosorbent assay. J Gen Microbiol132:43–52. https://doi.org/10.1099/00221287-132-1-43.

36. Margherita S, Hungate R. 1963. Serological analysis of Butyrivibrio fromthe bovine rumen. J Bacteriol 86:855– 860.

37. Palevich N. 2016. Comparative genomics of Butyrivibrio and Pseudobu-tyrivibrio from the rumen. PhD dissertation. Massey University, Palmer-ston North, New Zealand.

38. Kim M, Oh H-S, Park S-C, Chun J. 2014. Towards a taxonomic coherencebetween average nucleotide identity and 16S rRNA gene sequencesimilarity for species demarcation of prokaryotes. Int J Syst Evol Micro-biol 64:346 –351. https://doi.org/10.1099/ijs.0.059774-0.

39. Richter M, Rosselló-Móra R. 2009. Shifting the genomic gold standardfor the prokaryotic species definition. Proc Natl Acad Sci U S A 106:19126 –19131. https://doi.org/10.1073/pnas.0906412106.

40. Biegel E, Müller V. 2010. Bacterial Na�-translocating ferredoxin: NAD�

oxidoreductase. Proc Natl Acad Sci U S A 107:18138 –18142. https://doi.org/10.1073/pnas.1010318107.

41. Tremblay PL, Zhang T, Dar SA, Leang C, Lovley DR. 2012. The Rnf complexof Clostridium ljungdahlii is a proton-translocating ferredoxin: NAD� oxi-doreductase essential for autotrophic growth. mBio 4:e00406-12. https://doi.org/10.1128/mBio.00406-12.

42. Hackmann TJ, Firkins JL. 2015. Electron transport phosphorylation inrumen butyrivibrios: unprecedented ATP yield for glucose fermenta-tion to butyrate. Front Microbiol 6:622. https://doi.org/10.3389/fmicb.2015.00622.

43. Hespell R, Wolf R, Bothast R. 1987. Fermentation of xylans by Butyriv-ibrio fibrisolvens and other ruminal bacteria. Appl Environ Microbiol53:2849 –2853

44. Dehority B. 1973. Hemicellulose degradation by rumen bacteria. FedProc 32:1819 –1825.

45. Gupta VK, Chaudhari NM, Iskepalli S, Dutta C. 2015. Divergences ingene repertoire among the reference Prevotella genomes derived from

distinct body sites of human. BMC Genomics 16:153. https://doi.org/10.1186/s12864-015-1350-6.

46. McInerney JO, McNally A, O’Connell MJ. 2017. Why prokaryotes havepangenomes. Nat Microbiol 2:17040. https://doi.org/10.1038/nmicrobiol.2017.40.

47. Hao W, Golding GB. 2006. The fate of laterally transferred genes: life inthe fast lane to adaptation or death. Genome Res 16:636 – 643. https://doi.org/10.1101/gr.4746406.

48. Teather RM. 1982. Isolation of plasmid DNA from Butyrivibrio fibrisol-vens. Appl Environ Microbiol 43:298 –302.

49. Kelly WJ, Leahy SC, Altermann E, Yeoman CJ, Dunne JC, Kong Z,Pacheco DM, Li D, Noel SJ, Moon CD, Cookson AL, Attwood GT. 2010.The glycobiome of the rumen bacterium Butyrivibrio proteoclasticusB316T highlights adaptation to a polysaccharide-rich environment.PLoS One 5:e11942. https://doi.org/10.1371/journal.pone.0011942.

50. Palevich N. 2011. Genome sequencing of rumen bacteria involved inlignocellulose digestion. MSc thesis. Massey University, PalmerstonNorth, New Zealand.

51. Yeoman CJ, Kelly WJ, Rakonjac J, Leahy SC, Altermann E, Attwood GT.2011. The large episomes of Butyrivibrio proteoclasticus B316T havearisen through intragenomic gene shuttling from the chromosome tosmaller Butyrivibrio-specific plasmids. Plasmid 66:67–78. https://doi.org/10.1016/j.plasmid.2011.05.002.

52. Palevich N, Kelly WJ, Leahy SC, Altermann E, Rakonjac J, Attwood GT.2017. The complete genome sequence of the rumen bacterium Butyr-ivibrio hungatei MB2003. Stand Genomic Sci 12:72. https://doi.org/10.1186/s40793-017-0285-8.

53. Harrison PW, Lower RP, Kim NK, Young JP. 2010. Introducing thebacterial ‘chromid’: not a chromosome, not a plasmid. Trends Microbiol18:141–148. https://doi.org/10.1016/j.tim.2009.12.010.

54. Morrison M. 1996. Do ruminal bacteria exchange genetic material? J DairySci 79:1476–1486. https://doi.org/10.3168/jds.S0022-0302(96)76507-4.

55. Kav AB, Sasson G, Jami E, Doron-Faigenboim A, Benhar I, Mizrahi I.2012. Insights into the bovine rumen plasmidome. Proc Natl Acad SciU S A 109:5452–5457. https://doi.org/10.1073/pnas.1116410109.

56. Jain R, Rivera MC, Moore JE, Lake JA. 2003. Horizontal gene transferaccelerates genome innovation and evolution. Mol Biol Evol 20:1598 –1602. https://doi.org/10.1093/molbev/msg154.

57. Selinger L, Forsberg C, Cheng K-J. 1996. The rumen: a unique source ofenzymes for enhancing livestock production. Anaerobe 2:263–284.https://doi.org/10.1006/anae.1996.0036.

58. Cheng K-J, Costerton JW. 1980. The formation of microcolonies byrumen bacteria. Can J Microbiol 26:1104 –1113. https://doi.org/10.1139/m80-183.

59. Kalmokoff M, Teather R. 1997. Isolation and characterization of a bac-teriocin (butyrivibriocin AR10) from the ruminal anaerobe Butyrivibriofibrisolvens AR10: evidence in support of the widespread occurrence ofbacteriocin-like activity among ruminal isolates of B. fibrisolvens. ApplEnviron Microbiol 63:394 – 402.

60. Kalmokoff ML, Bartlett F, Teather RM. 1996. Are ruminal bacteria armedwith bacteriocins? J Dairy Sci 79:2297–2306. https://doi.org/10.3168/jds.S0022-0302(96)76608-0.

61. Flint HJ, Thomson AM, Bisset J. 1988. Plasmid-associated transfer oftetracycline resistance in Bacteroides ruminicola. Appl Environ Microbiol54:855– 860.

62. Shoemaker NB, Wang G-R, Salyers AA. 1992. Evidence for naturaltransfer of a tetracycline resistance gene between bacteria from thehuman colon and bacteria from the bovine rumen. Appl EnvironMicrobiol 58:1313–1320.

63. Palevich N, Kelly WJ, Ganesh S, Rakonjac J, Attwood GT. 2018. Butyrivibriohungatei MB2003 competes effectively for soluble sugars released byButyrivibrio proteoclasticus B316T during growth on xylan or pectin. ApplEnviron Microbiol 85:e02056-18. https://doi.org/10.1128/AEM.02056-18.

64. Welte C, Krätzer C, Deppenmeier U. 2010. Involvement of Ech hydro-genase in energy conservation of Methanosarcina mazei. FEBS J 277:3396 –3403. https://doi.org/10.1111/j.1742-4658.2010.07744.x.

65. Buckel W, Thauer RK. 2013. Energy conservation via electron bifurcatingferredoxin reduction and proton/Na� translocating ferredoxin oxidation.Biochim Biophys Acta 1827:94–113. https://doi.org/10.1016/j.bbabio.2012.07.002.

66. Gutekunst K, Chen X, Schreiber K, Kaspar U, Makam S, Appel J. 2014. Thebidirectional NiFe-hydrogenase in Synechocystis sp. PCC 6803 is reducedby flavodoxin and ferredoxin and is essential under mixotrophic, nitrate-

Polysaccharide-Degrading Capabilities of Butyrivibrio Applied and Environmental Microbiology

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 17

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from

limiting conditions. J Biol Chem 289:1930–1937. https://doi.org/10.1074/jbc.M113.526376.

67. Pereira IA, Ramos AR, Grein F, Marques MC, Da Silva SM, Venceslau SS.2011. A comparative genomic analysis of energy metabolism in sulfatereducing bacteria and archaea. Front Microbiol 2:69. https://doi.org/10.3389/fmicb.2011.00069.

68. Weghoff MC, Bertsch J, Müller V. 2015. A novel mode of lactatemetabolism in strictly anaerobic bacteria. Environ Microbiol 17:670 – 677. https://doi.org/10.1111/1462-2920.12493.

69. Cooper R. 1984. Metabolism of methylglyoxal in microorganisms. AnnuRev Microbiol 38:49 – 68. https://doi.org/10.1146/annurev.micro.38.1.49.

70. Kenters N, Henderson G, Jeyanathan J, Kittelmann S, Janssen PH. 2011.Isolation of previously uncultured rumen bacteria by dilution to extinc-tion using a new liquid culture medium. J Microbiol Methods 84:52– 60.https://doi.org/10.1016/j.mimet.2010.10.011.

71. Duthoit F, Godon J-J, Montel M-C. 2003. Bacterial community dynamicsduring production of registered designation of origin Salers cheese asevaluated by 16S rRNA gene single-strand conformation polymorphismanalysis. Appl Environ Microbiol 69:3840–3848. https://doi.org/10.1128/aem.69.7.3840-3848.2003.

72. Escalante A, Wacher C, Farrés A. 2001. Lactic acid bacterial diversity inthe traditional Mexican fermented dough pozol as determined by 16SrDNA sequence analysis. Int J Food Microbiol 64:21–31. https://doi.org/10.1016/s0168-1605(00)00428-1.

73. Lane DJ. 1991. 16S/23S rRNA sequencing, p 115–147. In Stackebrandt E,Goodfellow M (ed), Nucleic acid techniques in bacterial systematics.John Wiley & Sons, Chichester, United Kingdom.

74. Dorsch M, Stackebrandt E. 1992. Some modifications in the procedureof direct sequencing of PCR amplified 16S rDNA. J Microbiol Methods16:271–279. https://doi.org/10.1016/0167-7012(92)90017-X.

75. Bergmann GT, Bates ST, Eilers KG, Lauber CL, Caporaso JG, Walters WA,Knight R, Fierer N. 2011. The under-recognized dominance of Verruco-microbia in soil bacterial communities. Soil Biol Biochem 43:1450 –1455.https://doi.org/10.1016/j.soilbio.2011.03.012.

76. Staden R, Beal KF, Bonfield JK. 1999. The Staden package. Methods MolBiol 132:115–130.

77. Kearse M, Moir R, Wilson A, Stones-Havas S, Cheung M, Sturrock S,Buxton S, Cooper A, Markowitz S, Duran C, Thierer T, Ashton B, MeintjesP, Drummond A. 2012. Geneious Basic: an integrated and extendabledesktop software platform for the organization and analysis of se-quence data. Bioinformatics 28:1647–1649. https://doi.org/10.1093/bioinformatics/bts199.

78. Pruitt KD, Tatusova T, Maglott DR. 2005. NCBI Reference Sequence(RefSeq): a curated non-redundant sequence database of genomes,transcripts and proteins. Nucleic Acids Res 33:D501–D504. https://doi.org/10.1093/nar/gki025.

79. Johnson M, Zaretskaya I, Raytselis Y, Merezhuk Y, McGinnis S, MaddenTL. 2008. NCBI BLAST: a better web interface. Nucleic Acids Res 36:W5–W9. https://doi.org/10.1093/nar/gkn201.

80. Maidak BL, Cole JR, Lilburn TG, Parker CT, Jr, Saxman PR, Farris RJ,Garrity GM, Olsen GJ, Schmidt TM, Tiedje JM. 2001. The RDP-II (Ribo-somal Database Project). Nucleic Acids Res 29:173–174. https://doi.org/10.1093/nar/29.1.173.

81. Thompson JD, Gibson T, Higgins DG. 2003. Multiple sequence align-ment using ClustalW and ClustalX. Curr Protoc Bioinformatics Chapter2:Unit 2.3.

82. Tamura K, Stecher G, Peterson D, Filipski A, Kumar S. 2013. MEGA6:Molecular Evolutionary Genetics Analysis version 6.0. Mol Biol Evol30:2725–2729. https://doi.org/10.1093/molbev/mst197.

83. Saitou N, Nei M. 1987. The neighbor-joining method: a new method forreconstructing phylogenetic trees. Mol Biol Evol 4:406 – 425. https://doi.org/10.1093/oxfordjournals.molbev.a040454.

84. Kimura M. 1980. A simple method for estimating evolutionary rates ofbase substitutions through comparative studies of nucleotide se-quences. J Mol Evol 16:111–120. https://doi.org/10.1007/bf01731581.

85. Akaike H. 1998. Information theory and an extension of the maximumlikelihood principle, p 199 –213. In Selected papers of Hirotugu Akaike.Springer, New York, NY.

86. Tittsler RP, Sandholzer LA. 1936. The use of semi-solid agar for thedetection of bacterial motility. J Bacteriol 31:575–580.

87. Attwood GT, Reilly K, Patel B. 1996. Clostridium proteoclasticum sp. nov.,a novel proteolytic bacterium from the bovine rumen. Int J Syst Bac-teriol 46:753–758. https://doi.org/10.1099/00207713-46-3-753.

88. Avguštin G, Wright F, Flint HJ. 1994. Genetic diversity and phylogeneticrelationships among strains of Prevotella (Bacteroides) ruminicola fromthe rumen. Int J Syst Bacteriol 44:246 –255. https://doi.org/10.1099/00207713-44-2-246.

89. Iverson WG, Millis NF. 1976. Characterization of Streptococcus bovisbacteriophages. Can J Microbiol 22:847– 852. https://doi.org/10.1139/m76-122.

90. Shimoyama T, Kato S, Ishii S, Watanabe K. 2009. Flagellum mediatessymbiosis. Science 323:1574. https://doi.org/10.1126/science.1170086.

91. Chen I-M, Markowitz VM, Chu K, Palaniappan K, Szeto E, Pillay M, RatnerA, Huang J, Andersen E, Huntemann M. 2017. IMG/M: integrated ge-nome and metagenome comparative data analysis system. NucleicAcids Res 45:D507–D516. https://doi.org/10.1093/nar/gkw929.

92. Altschul SF, Madden TL, Schäffer AA, Zhang J, Zhang Z, Miller W,Lipman DJ. 1997. Gapped BLAST and PSI-BLAST: a new generation ofprotein database search programs. Nucleic Acids Res 25:3389 –3402.https://doi.org/10.1093/nar/25.17.3389.

93. Richardson A, Calder A, Stewart C, Smith A. 1989. Simultaneous deter-mination of volatile and non-volatile acidic fermentation products ofanaerobes by capillary gas chromatography. Lett Appl Microbiol 9:5– 8.https://doi.org/10.1111/j.1472-765X.1989.tb00278.x.

94. White JR, Nagarajan N, Pop M. 2009. Statistical methods for detectingdifferentially abundant features in clinical metagenomic samples. PLoSComput Biol 5:e1000352. https://doi.org/10.1371/journal.pcbi.1000352.

95. Mavromatis K, Land ML, Brettin TS, Quest DJ, Copeland A, Clum A,Goodwin L, Woyke T, Lapidus A, Klenk HP, Cottingham RW, KyrpidesNC. 2012. The fast changing landscape of sequencing technologies andtheir impact on microbial genome assemblies and annotation. PLoSOne 7:e48837. https://doi.org/10.1371/journal.pone.0048837.

96. Eid J, Fehr A, Gray J, Luong K, Lyle J, Otto G, Peluso P, Rank D, BaybayanP, Bettman B, Bibillo A, Bjornson K, Chaudhuri B, Christians F, Cicero R,Clark S, Dalal R, Dewinter A, Dixon J, Foquet M, Gaertner A, HardenbolP, Heiner C, Hester K, Holden D, Kearns G, Kong X, Kuse R, Lacroix Y, LinS, Lundquist P, Ma C, Marks P, Maxham M, Murphy D, Park I, Pham T,Phillips M, Roy J, Sebra R, Shen G, Sorenson J, Tomaney A, Travers K,Trulson M, Vieceli J, Wegener J, Wu D, Yang A, Zaccarin D, et al. 2009.Real-time DNA sequencing from single polymerase molecules. Science323:133–138. https://doi.org/10.1126/science.1162986.

97. Zerbino DR, Birney E. 2008. Velvet: algorithms for de novo short readassembly using de Bruijn graphs. Genome Res 18:821– 829. https://doi.org/10.1101/gr.074492.107.

98. Butler J, MacCallum I, Kleber M, Shlyakhter IA, Belmonte MK, Lander ES,Nusbaum C, Jaffe DB. 2008. ALLPATHS: de novo assembly of whole-genome shotgun microreads. Genome Res 18:810 – 820. https://doi.org/10.1101/gr.7337908.

99. Chin C-S, Alexander DH, Marks P, Klammer AA, Drake J, Heiner C, ClumA, Copeland A, Huddleston J, Eichler EE, Turner SW, Korlach J. 2013.Nonhybrid, finished microbial genome assemblies from long-readSMRT sequencing data. Nat Methods 10:563–569. https://doi.org/10.1038/nmeth.2474.

100. Huntemann M, Ivanova NN, Mavromatis K, Tripp HJ, Paez-Espino D,Tennessen K, Palaniappan K, Szeto E, Pillay M, Chen I-MA, Pati A,Nielsen T, Markowitz VM, Kyrpides NC. 2016. The standard operatingprocedure of the DOE-JGI Metagenome Annotation Pipeline (MAP v. 4).Stand Genomic Sci 11:17. https://doi.org/10.1186/s40793-016-0138-x.

101. Tripp HJ, Sutton G, White O, Wortman J, Pati A, Mikhailova N, Ovchin-nikova G, Payne SH, Kyrpides NC, Ivanova N. 2015. Toward a standardin structural genome annotation for prokaryotes. Stand Genomic Sci10:45. https://doi.org/10.1186/s40793-015-0034-9.

102. Hyatt D, Chen G-L, LoCascio PF, Land ML, Larimer FW, Hauser LJ. 2010.Prodigal: prokaryotic gene recognition and translation initiation siteidentification. BMC Bioinformatics 11:119. https://doi.org/10.1186/1471-2105-11-119.

103. Pati A, Ivanova NN, Mikhailova N, Ovchinnikova G, Hooper SD, LykidisA, Kyrpides NC. 2010. GenePRIMP: a gene prediction improvementpipeline for prokaryotic genomes. Nat Methods 7:455– 457. https://doi.org/10.1038/nmeth.1457.

104. Yin Y, Mao X, Yang J, Chen X, Mao F, Xu Y. 2012. dbCAN: a web resourcefor automated carbohydrate-active enzyme annotation. Nucleic AcidsRes 40:W445–W451. https://doi.org/10.1093/nar/gks479.

105. Marchler-Bauer A, Zheng C, Chitsaz F, Derbyshire MK, Geer LY, Geer RC,Gonzales NR, Gwadz M, Hurwitz DI, Lanczycki CJ, Lu F, Lu S, MarchlerGH, Song JS, Thanki N, Yamashita RA, Zhang D, Bryant SH. 2013. CDD:

Palevich et al. Applied and Environmental Microbiology

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 18

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from

conserved domains and protein three-dimensional structure. NucleicAcids Res 41:D348 –D352. https://doi.org/10.1093/nar/gks1243.

106. Cantarel BL, Coutinho PM, Rancurel C, Bernard T, Lombard V, HenrissatB. 2009. The Carbohydrate-Active EnZymes database (CAZy): an expertresource for glycogenomics. Nucleic Acids Res 37:D233–D238. https://doi.org/10.1093/nar/gkn663.

107. Finn RD, Bateman A, Clements J, Coggill P, Eberhardt RY, Eddy SR,Heger A, Hetherington K, Holm L, Mistry J, Sonnhammer ELL, Tate J,Punta M. 2014. Pfam: the protein families database. Nucleic Acids Res42:D222–D230. https://doi.org/10.1093/nar/gkt1223.

108. Rose PW, Bi C, Bluhm WF, Christie CH, Dimitropoulos D, Dutta S, GreenRK, Goodsell DS, Prlic A, Quesada M, Quinn GB, Ramos AG, WestbrookJD, Young J, Zardecki C, Berman HM, Bourne PE. 2013. The RCSB ProteinData Bank: new resources for research and education. Nucleic Acids Res41:D475–D482. https://doi.org/10.1093/nar/gks1200.

109. Kaznadzey A, Alexandrova N, Novichkov V, Kaznadzey D. 2013.PSimScan: algorithm and utility for fast protein similarity search. PLoSOne 8:e58505. https://doi.org/10.1371/journal.pone.0058505.

110. Altermann E. 2012. Tracing lifestyle adaptation in prokaryotic genomes.Front Microbiol 3:48. https://doi.org/10.3389/fmicb.2012.00048.

111. Markowitz VM, Chen I-MA, Palaniappan K, Chu K, Szeto E, Pillay M,Ratner A, Huang J, Woyke T, Huntemann M, Anderson I, Billis K,Varghese N, Mavromatis K, Pati A, Ivanova NN, Kyrpides NC. 2014. IMG4 version of the integrated microbial genomes comparative analysissystem. Nucleic Acids Res 42:D560 –D567. https://doi.org/10.1093/nar/gkt963.

112. Altermann E, Lu J, McCulloch A. 2017. GAMOLA2, a comprehensivesoftware package for the annotation and curation of draft and com-plete microbial genomes. Front Microbiol 8:346. https://doi.org/10.3389/fmicb.2017.00346.

113. Sneath PH, Sokal RR. 1962. Numerical taxonomy. Nature 193:855– 860.https://doi.org/10.1038/193855a0.

114. Jones DT, Taylor WR, Thornton JM. 1992. The rapid generation ofmutation data matrices from protein sequences. Comput Appl Biosci8:275–282. https://doi.org/10.1093/bioinformatics/8.3.275.

115. Delcher AL, Kasif S, Fleischmann RD, Peterson J, White O, Salzberg SL.1999. Alignment of whole genomes. Nucleic Acids Res 27:2369 –2376.https://doi.org/10.1093/nar/27.11.2369.

116. Womble DD. 1999. GCG: The Wisconsin package of sequence analysisprograms. Methods Mol Biol 132:3–22.

117. Vesth T, Lagesen K, Acar O, Ussery D. 2013. CMG-biotools, a freeworkbench for basic comparative microbial genomics. PLoS One8:e60120. https://doi.org/10.1371/journal.pone.0060120.

118. Ihaka R, Gentleman R. 1996. R: a language for data analysis andgraphics. J Comput Graph Stat 5:299 –314. https://doi.org/10.2307/1390807.

119. Bray JR, Curtis JT. 1957. An ordination of the upland forest communitiesof southern Wisconsin. Ecol Monogr 27:325–349. https://doi.org/10.2307/1942268.

120. Ward JH. 1963. Hierarchical grouping to optimize an objective function.J Am Stat Assoc 58:236 –244. https://doi.org/10.2307/2282967.

121. Caspi R, Altman T, Dreher K, Fulcher CA, Subhraveti P, Keseler IM,Kothari A, Krummenacker M, Latendresse M, Mueller LA, Ong Q, PaleyS, Pujar A, Shearer AG, Travers M, Weerasinghe D, Zhang P, Karp PD.2012. The MetaCyc database of metabolic pathways and enzymes andthe BioCyc collection of pathway/genome databases. Nucleic Acids Res40:D742–D753. https://doi.org/10.1093/nar/gkr1014.

122. Kanehisa M, Goto S. 2000. KEGG: Kyoto encyclopedia of genes and ge-nomes. Nucleic Acids Res 28:27–30. https://doi.org/10.1093/nar/28.1.27.

Polysaccharide-Degrading Capabilities of Butyrivibrio Applied and Environmental Microbiology

January 2020 Volume 86 Issue 1 e01993-19 aem.asm.org 19

on October 13, 2020 by guest

http://aem.asm

.org/D

ownloaded from